Mechanism and data hybrid driven generative adversarial network soft measurement modeling method

A mechanism and data hybrid driven generative adversarial network soft measurement modeling method comprises the following steps: sampling from an auxiliary variable to obtain data q, sampling from random noise to obtain data z, inputting q and z into a to-be-identified generator composed of a mechanism model and a data driven error compensation model to obtain a generated sample, inputting the generated sample and a real sample x into a discriminator, and performing back propagation on the obtained loss to obtain an optimal generator parameter; taking out the optimal generator as a trained soft measurement model, analyzing the interpretability of the model by using a latent variable manipulation method, and analyzing the prediction performance of the model by comparing with a single-drive soft measurement model; the method overcomes the limitations that a static model cannot capture dynamic characteristics among data, a mechanism-driven soft measurement model has modeling errors, data-driven soft measurement